Google Assistant

Google Assistant is a program by the company Google that you can talk to. You say a sentence, the program understands it and reacts accordingly. It answers questions, sets alarms, plays music, or turns off the light in the living room. It is usually started with the words “Hey Google” or by pressing a button for a long time. It was introduced in 2016 and has since run on phones with Google’s Android operating system, on small speakers, and on watches. Such programs are called voice assistants; well-known competitors are Siri by Apple and Alexa by Amazon.

Speech as a user interface

Voice assistants have changed the way people use computers. Before, you always needed a screen and fingers or a mouse. With an assistant, a sentence from the next room is enough. This is practical while cooking, while driving, or for people who have poor eyesight or find it hard to press buttons.

For Google itself, though, there was more at stake. Anyone who no longer types but asks also no longer sees a results list with ads. The assistant delivers only a single answer. This puts at risk the business model of the classic search engine, which for years brought in almost all of Google’s money. That is exactly why Google, Apple, and Amazon invested heavily in this technology starting in 2015.

A second point is the smart home, meaning lamps, heaters, and power outlets that can be controlled over the internet. Whoever controls the assistant controls the remote control for the entire home. Manufacturers of such devices therefore have to stay on good terms with Google.

From microphone to answer

The process consists of several steps. First, the device only listens for the wake word “Hey Google.” This is handled by a very small program directly on the device, without internet access. Only after that does the actual recording begin. The audio is usually sent to Google’s data centers and converted into text there. This step is called speech recognition.

After that, the intent must be recognized. “Set an alarm for seven” and “wake me up at seven” mean the same thing but sound different. The system assigns the sentence to a task and extracts the important information, in this case the time. It then carries out the task and formulates a response, which is played back as an artificial voice.

A common misconception is that the device constantly records everything and sends it away. Normally, nothing leaves the device before the wake word. Still, there was criticism: in 2019 it became known that employees listened to individual recordings in order to improve recognition. False activations also occur when a similar-sounding sentence wakes the device.

On the phone, in the speaker – and the switch to Gemini

The assistant is most commonly encountered on Android phones, where it is deeply embedded in the system. In addition, there are the speakers and displays of the Google Home and Nest lines, TVs with Android TV, headphones, smartwatches, and cars with Android Auto. Other manufacturers also build it in, for example in lamps or thermostats.

In the news, the name has mainly come up since 2024 in connection with the overhaul. Google is gradually replacing the assistant with Gemini, its newer AI system, which can hold entire conversations and formulate texts on its own. The old assistant relied heavily on fixed, pre-programmed commands, whereas Gemini responds more freely. The switch has been rolling out on phones and speakers since 2025.

For users, this means more possibilities, but also an adjustment period. Some familiar functions disappear or are renamed. For the industry, the overhaul is a signal: voice assistants of the old design are considered outdated, because modern language models solve the same task more flexibly.